# 1、下载代码

git clone github.com/wxj630/visual-chatgpt-zh

# 2、进入项目目录
cd visual-chatgpt-zh

# 3、创建python环境

conda create -n visgpt python=3.8

# 4、安装环境依赖

pip install -r requirement.txt   -i https://pypi.douban.com/simple

安装basicsr时 有

pip._vendor.urllib3.exceptions.ReadTimeoutError: HTTPSConnectionPool(host='files.pythonhosted.org', port=443): Read timed out

通过代理解决，python使用代理时需要先安装pysocks库

pip install pysocks

set all_proxy=socks5://192.168.0.10:7070
set https_proxy=socks5://192.168.0.10:7070
set https_proxy=socks5://192.168.0.10:7070

# 5、确认api key
export OPENAI_API_KEY={Your_Private_Openai_Key}

# 6、下载hf模型到指定目录（注意要修改sh文件里的{your_hf_models_path}为模型存放目录）
bash download_hf_models.sh

# 7、克隆ContrlNet的代码，建立软链接，并下载ControlNet需要的模型
bash download_controlnet_models.sh

这个下载需要很久，下载模型也有点大

win下 download_hf_models.bat

git clone https://huggingface.co/Salesforce/blip-image-captioning-base
git clone https://huggingface.co/runwayml/stable-diffusion-v1-5
git clone https://huggingface.co/runwayml/stable-diffusion-inpainting
git clone https://huggingface.co/CIDAS/clipseg-rd64-refined
git clone https://huggingface.co/timbrooks/instruct-pix2pix
git clone https://huggingface.co/Salesforce/blip-vqa-base
git clone https://huggingface.co/lllyasviel/ControlNet
git clone https://huggingface.co/lllyasviel/sd-controlnet-canny 
git clone https://huggingface.co/lllyasviel/sd-controlnet-seg 
git clone https://huggingface.co/lllyasviel/sd-controlnet-scribble 
git clone https://huggingface.co/lllyasviel/sd-controlnet-normal 
git clone https://huggingface.co/lllyasviel/sd-controlnet-mlsd 
git clone https://huggingface.co/lllyasviel/sd-controlnet-depth 
git clone https://huggingface.co/lllyasviel/sd-controlnet-hed 
git clone https://huggingface.co/lllyasviel/sd-controlnet-openpose 
git clone https://huggingface.co/openmmlab/upernet-convnext-small 
git clone https://huggingface.co/Intel/dpt-hybrid-midas

# 8、创建一个文件夹存放图片
mkdir ./image

# 9、启动系统！（注意要修改sh文件里的{your_hf_models_path}为模型存放目录）
sh run.sh

https://www.panewslab.com/zh/articledetails/kz49zxoytl9e.html

https://zhuanlan.zhihu.com/p/613449915

https://github.com/microsoft/visual-chatgpt
https://github.com/wxj630/visual-chatgpt-zh

Visual ChatGPT 视觉模型深度解析

By:xSky | 人工智能/机器学习 |